Technology for haptic interfaces

Haptic Interfaces are essential components of Enactive Interfaces. Indeed this component is conceptually the only possible vector of the Human action toward virtual objects or environments. One of the activities of the Network is focused on the analysis of the existing technology in order to identify its limits and potentialities for the needs of Enaction. From this analysis several experiments have been carried out or are planned. They will allow focusing analysis on identified critical points and validating some new technological improvements. In close relationship with these technological experimentations new haptic devices and new important haptic component have been realized or designed. These new devices will permit more efficient approaches of Enaction either in the design and realization of Enactive interfaces for usages or for scientific experimentations in order to achieve a more accurate understanding of this concept.
The following describes the investigations and results concerning technological analysis, experimentations and realizations of new haptic devices related to the main critical aspects of today’s haptic interfaces: spatial issues, dynamical issues and issues related to the direct contact and the integration of tactile /force feedback.

Spatial issues - Large workspaces - Torque feedback

The lack of haptic devices able to show both high mechanical performances and large workspaces was highlighted in the Roadmap Document. This issue was strongly related to the lack of light weight electro-mechanical actuators. At PERCRO Lab a new prototype of an integrated structural, actuation and speed reduction system was developped. The developed system (see picture below) transforms the rotational motion of an electrical motor into a liner movement making use of a ball bearing screw. The linear motion is changed back into a rotation thanks to a mechanism specially developed for this purpose.

Fig 1. Prototype of an integrated structural actuation and speed reduction system.

Another important spatial issue concerns the restitution of torques in haptic interfaces. The issues is investigated by CEIT through specific experiments.
These experiments have been carried out on a desk-grounded 3-dof wrist from CEIT laboratory. It is a parallel 2-dof orientation mechanism with a serially coupled hand-roll dof along the handle axis. Parallel dofs are driven by commercial Maxon motors and specially designed and built cable transmissions. The hand-roll is driven by a Maxon motor with a Maxon planetary gearhead. Rotation of each dof is measured by an encoder coupled to its motor. By design, rotation of the handle around its axis can be set to a fixed position, and this way the device becomes a 2-dof wrist. The handle is also provided with a force/torque sensor to measure the torque exerted by the user along its axis when holding it. Rotation of the handle around its axis is fixed throughout the trials, so the device behaves as a 2-dof wrist, lacking of the hand-roll dof.
The device is controlled by a dSPACE DS1104 board that reads encoder information, processes the haptic control loop and outputs torque commands to the motors. Graphic rendering and collision detection are performed in a Pentium IV 2.5 GHz PC with 512 MB RAM, NVIDIA GeForce 4 Ti 4200 graphics card and Windows XP operating system. The overall specifications of the interface are resumed in the above table.
An evaluation based on different users’ comments shows that, to a certain extent, all of them have a perception of 3 dof even if mechanically limited to 2 dof. Real haptic dofs and pseudo-haptic dofs have been successfully combined into a unique device with the visual strategies proposed. It has been observed that, thanks to the visual dominance on human perception, terminal cues are suitably addressed by visual feedback to help the user perceive collisions.

Fig. 2. 3-DoF torque feedback device, convertible to 2-DoF

Dynamical issues. High bandwidth Haptic devices & Timing limitations due to OS and data Busses.

One of the bottlenecks of haptic applications is related to the limited frequency at which the simulation loop is run (usually between 1 and 3 kHz). A reduced frequency of the simulation induces for example lower stability and a smaller range of possible simulation parameters.
Given the very high dynamical characteristics of the ERGOS haptic device, INPG found interesting to benefit from the possibility of generating very fast simulation loops in real time. The main advantages of high frequency simulation are:

1. To increase the overall stability and parameter consistency of the simulated model
2. The possibility to provide harder contacts, and to increase the maximum stiffness available
3. To compute in the same simulation loop both the haptic and the audio parts of the model in real time (actually within one single model), without any multi-frequency workaround features. Avoiding such features leads to simplify the model and enhances again the benefit of the fast computing rate.
4. To use two of the D/A channels to feed directly the sound output from the processing unit to the loudspeakers, thus avoiding the filtering delay due to the of additional peripherals for sound output. The 16bits resolution of the D/A converters, however small considering the mainstream today in terms of resolution for audio output, was found to be sufficient for such a use.
1. The first interesting application of running simulations at very high frequencies is to design simulations where sound and haptics are computed within the same simulation loop. When the sound is generated immediately at the output of the D/A converters, it is necessary to have the simulation loop running above 20-25 kHz, so that the effects of digitization cannot be perceived by human ear. Moreover, this technique allows to build applications where the haptic and the audio parts are intimately coupled, as for real sounding objects.
2. Increasing the frequency of the simulation loop is of great interest for pure haptic applications too. Using a suitable haptic device capable of very high dynamical properties (such as the ERGOS haptic electromagnetic device), it is possible to raise a new level of stiffness and damping that was unreachable up to now with usual haptic devices. It constitutes an innovative technological step that opens new perspectives in the field of believable materials and rigid objects.

Fig. 3. Corner-in-the-box application : the simulation of very hard contacts is made possible thanks to the high bandwidth Haptic Interface Ergos combined with the running of the simulation loop at a very high frequency (40 kHz).

Haptic devices still lack of plug’n play features, which would make easier the exchange of materials and the spreading of haptic uses. The USB communication protocol is currently one of the most widely used technique for the connection of computer peripherals.
A series of experiments has been executed with the DLR USB force feedback Joystick using the rubber band demonstration under different operating systems and platforms. The aim was to identify timing and stability errors of the USB connection and if possible timing differences (benchmarking). The rubber band demonstration gives us the possibility to bounce a virtual ball over a simulated rubber band in an environment with adaptable damping. The mass of the ball can be changed between 0 and 100% (14Nm torque at 150 mm length of the joystick handle). A virtual wall with contact possibilities in both joystick axes gives the quality impression of hard contacts in both directions of the joystick.

Fig 4. DLR’s USB Force feedback joystick

Direct CONTACT ISSUE and Haptic Tactile integration.

Experiment related to the lack of direct contact

The lack of direct contact haptic devices was one of the most critical points highlighted in the Roadmap report. Some experiments were set up at PERCRO Lab in order to explore some effective solutions for the simulation of direct contact interaction with virtual objects. In particular our experiments focus on the identification of different solutions for the mechanical connection between the user and the haptic device. We realize several prototypes and we carried on some early tests. The different adopted solutions are composed of prototypes for the improvement of lateral force perception and Contact Transient simulation (see figure below).

Fig. 5. Improvement of the lateral force perception

PERCRO also carried out other developments related to the improvement of the accuracy of force feedback device and to the integration of tactile and haptic feedbacks. The GRAB system was modified by adding a force sensor at the end-effector and by adding the angular sensorization on the last 3 DoF.
The new end-effector was designed in order to host a Tactile Array developed by UNEXE - University of Exeter.(see picture below)

Fig 7. — Design of the new end-effector of the Grab in order to host a Tactile Array developed by the university of EXETER

Similar works related to the integration of a tactile stimulator to different force feedback devices have been carried on at COSTECH and INPG.
The main novelty such device is that the tactile stimulation is obtained strictly from the same interaction loop, and obeys to the same physical formalism, as the FF. Thus, it provides both information on the spatial distribution of forces circulating between the object and the body (activation of tactile pins); and also permits to implement the deformable body.

Fig 8. Two tactile-force-feedback device: TactPHANToM (COSTECH, on the left) TactERGOS (INPG-COSTECH, on the right)

Still in the aim of integrating a tactile feedback with the force feedback, experiments have been performed at the University of Exeter to investigate discrimination of virtual textures produced by mixtures of sinewaves, and differentiated in terms of the mean amplitude and the spatial distribution at each sinewave frequency. The tactile stimulator used for this study (Figure 9) is an evolution of an earlier system developed by UNEXE.
The 2D workspace is available for free, active exploration by the test subject. It contains three squared targets, each 35 mm x 35 mm. When the tactile cursor lies outside the target areas, no vibratory stimulation is delivered to the fingertip. When the cursor lies within a target, the fingertip is presented with a texture. Perhaps the most interesting result is a strong interaction between the perceived spatial aspects of the texture and the stimulation frequency. If the stimulation frequency is changed from 40 Hz to 320 Hz, the perceived sensation during active exploration changes much more if the texture is spatially non-uniform than if it is spatially uniform.

Figure 9. Top: the tactile stimulator. In the inset screen shot, the tactile cursor, which represents the stimulated area, is visible between two of the square targets. Bottom: exploratory path – subjects were asked to explore at approximately constant speed – around 10 seconds for a single pass